Open-Source AI Models: Economics, Global Competition, and Cybersecurity Risks

TechCrunch's Equity podcast hosts Irene Solaiman, Chief Policy Officer and Head of Strategy at Hugging Face, to examine how open-weight AI models are reshaping the industry. The discussion covers how companies can generate revenue from freely available models, how Chinese models are competing for global adoption, and how security incidents affect the open ecosystem.

■ The Business of Open Models
- Monetization paths for model providers: inference costs, cloud revenue sharing, and custom licensing terms
- Bundling strategies used by Chinese providers versus model-only US players
- Enterprise and sovereign deployment deals as an alternative revenue route

■ Choosing and Building with Open Models
- Matching model capability and cost to specific tasks, and the role of inference budgets
- The importance of harnesses and tooling around models, plus the current gap in their evaluation
- GPU purchasing versus renting, local deployment, and fine-tuning resources on Hugging Face

■ Global Competition and Security
- Demand for US open models such as GPT-OSS alongside Chinese models like Qwen in sovereign AI projects
- Cybersecurity incidents involving Hugging Face and Open Code, and the trade-off between openness and safety
- The debate over unfiltered models, defense research, and incident-reporting standards

Ideal for founders, developers, and policy or security professionals evaluating open-source AI. It offers a concrete overview of cost, competition, and risk considerations to weigh before building on open weights.